ReviewRobot, from Wang et al. (2020), generates explainable paper reviews by building knowledge graphs from the target paper and related work, producing structured review comments with reasons attached.
Benefits & outcomes
71.4%–100% accuracy in review score predictions.
Human assessment found 41.7%–70.5% of generated comments valid and constructive.
Notes & quotes
"ReviewRobot can serve as an assistant for paper reviewers, program chairs and authors."
Publisher / company: Research Team (Wang et al.)
Date of mention: 01/10/2020
Source: https://arxiv.org/abs/2010.06119